学习主题，而非语言：如何LLMs在跨语言环境下分类在线移民 discourse
Learning the Topic, Not the Language: How LLMs Classify Online Immigration Discourse Across Languages
AI总结：
本文提出一种轻量级开源框架，利用微调后的LLaMA模型跨13种语言分类移民相关推文，展示LLMs在少量数据下可泛化主题理解，并通过多语言微调捕捉意识形态细微差别，实现高效且低成本的多语言分析。
摘要：
大型语言模型（LLMs）为在线 discourse 的可扩展分析提供了新的机会。然而，它们在多语言社会科学研究中的应用仍受到模型大小、成本和语言偏见的限制。我们开发了一个轻量级、开源的LLM框架，使用微调后的LLaMA 3.2-3B模型来跨13种语言分类与移民相关的推文。与以往依赖BERT风格模型或翻译管道的方法不同，我们结合了主题分类与立场检测，并证明了仅在一种或两种语言上微调的LLMs能够泛化到未见过的语言。然而，捕捉意识形态细微差别则受益于多语言微调。我们的方法通过少量来自代表性不足语言的数据纠正预训练偏见，并避免依赖专有系统。与商业LLMs相比，我们的方法在推理速度上快26-168倍，成本节省超过1000倍，支持实时分析数十亿条推文。这种以规模优先的框架使能够在语言和文化背景下进行包容、可重复的研究，研究公众态度。
英文摘要：
Large language models (LLMs) offer new opportunities for scalable analysis of online discourse. Yet their use in multilingual social science research remains constrained by model size, cost and linguistic bias. We develop a lightweight, open-source LLM framework using fine-tuned LLaMA 3.2-3B models to classify immigration-related tweets across 13 languages. Unlike prior work relying on BERT style models or translation pipelines, we combine topic classification with stance detection and demonstrate that LLMs fine-tuned in just one or two languages can generalize topic understanding to unseen languages. Capturing ideological nuance, however, benefits from multilingual fine-tuning. Our approach corrects pretraining biases with minimal data from under-represented languages and avoids reliance on proprietary systems. With 26-168x faster inference and over 1000x cost savings compared to commercial LLMs, our method supports real-time analysis of billions of tweets. This scale-first framework enables inclusive, reproducible research on public attitudes across linguistic and cultural contexts.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
